Thanks to visit codestin.com
Credit goes to www.scribd.com

0% found this document useful (0 votes)
20 views22 pages

Binomial, Poisson and Normal Distribution

1) A discrete probability distribution specifies the possible values and probabilities of a random variable. 2) Values are mutually exclusive and the sum of the probabilities is 1. 3) The probability mass function defines the discrete probability distribution and gives the probability that the random variable equals each possible value.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
20 views22 pages

Binomial, Poisson and Normal Distribution

1) A discrete probability distribution specifies the possible values and probabilities of a random variable. 2) Values are mutually exclusive and the sum of the probabilities is 1. 3) The probability mass function defines the discrete probability distribution and gives the probability that the random variable equals each possible value.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

AND PROBABILITY

,;,_:'l',, :;,:...-;~~f..:.; :,;-:,.~~:-f.:;:c:0r.. (o,: ~ir;-1civ Distribution) of a random variable is a Stareli:ffil


:,::r--+·=
-
·:.· .~,_.,_.,.- e Jc.,_
- .
- .. _,_,., .. ,~, -··
:" ·::::. ~ .... ,-.-:= r:-~ --.o,,i};J..,.
. . ,,,. ,_ ,._,~ _____ !J - ~--- Ir.res roo-;,,th.::,.r
f~ ='""' '- with th.::o.ir
"'- re-rrctivc
.:!)t'_ probabiliric.
-:;>-~.-:::::::.: ~- :"';:::-:d;);r_ e:tpe;frnent is theoretically assumed to serve as a model. dif
;..,,, .. ,,, i-;;,_;-.;·;·_;·_
-r· .,.,,, .. ,,,, .. .. ._.....:.- .... , Ct=--
_ ~.,, .::..2;- ~ ~J-·er:
- ..:::.
~ .,"""" _, -;.-., ~ ru
·· · rn "--1
...r=or.,J of th.:,."" random ran·ao' I-=- The probabiliF
y \.... •

; ;;..~_: ·_ - - ~-,.. .,._.._;:


..•-··-=----=--f
. , •. -~,.... : . ;,,. . ._ . .. j _ .
:. r~--
. i...;..✓• ~-v- l~ - - ·-all.
..:.ten genc::1 j kn
• oi;..n as- rJzeorenca
·.· [ azsm
,. "b ul!on.
.- -n: ,~-1n
nc ~
~:~✓..;..: ...~..;. ::~~ i-.ei~=-;,r~,-, frn::r:-- rm:::or~nt
. th,eoretical distributions.. ,·iz. Binomial. Po~SIJG..

-~ nr~J"_;P...&TE PROBABILITY DISTRIBUTION


L~ ~- d~:1.1~'::;t:; r~&~rri r;arfabie X a.~.surne the values x 1, x.,~ ... , x,1 \vith probabiliri6 _f1•
J;,-::; n.-• P,~-r~p,:..:-:ti;,·,;:;iy! where r,pr = l. The specification ot- the set of valu~s xi cog,_~
?",1f-"_:., t'i.ei:- ~rr.;.f>~;iE:fe~ P! (i = 1 2: ---: nJ defines the probability distribution ot roe
!

d;v.ztt...e r~-:.{~(;;'(( ,,·arfa~1'.~ Y.r {Jr in ~ho~ discrete probabilit;-· discriburioll of X.

Tabla 12 . 1 Di.wr&te Probabil it-.v Distribution.......,. ;.,


.
rn eo re tic al Di str ib ut io ns -B ino mi al , Poi
sso n, No rm al 42 1

. f!f.JII
r) is known, it is unnecessary tow nte the p b ... s p P b
/(
Jf · . f (x) on pu ttin g x _ . . ro ob,htie ) 1' 2• ...,p,,, ecauscp.
be obt ain ed from
· h d fi d · ·
- -\;, I.e., P· -Jc X; . he discrete probabTr '
T
piB}'
me sm 1pJ y by stating th ' -
·«ribulion is t en e . e mathema
x x 2, ... , x,,. tical expression fior f,(,x)y,
J1. with the set of po ssible values 1, , . .
,iong
,
Th eo re tic al Ois .
Table 12 • ·2~~ ~ .~.T-~ . tr1 · n (D.isc re te) ·
bu t io
- . -:,.•. s:.r.uxx.c:c:♦.,
---d dpl l!.- ir.:~,-a.i: .111 11
~ r ~- .Qg Jl!~ a.s:
X2 ~11a
X X1 .. J x,, Total

p=f(x)

The discrete ran dom variable •X may also assume countably infinite number of
. . .
p.m.f.J(x).
Possible value~ ~11 .t2, ... , .x,,, .. : wtth .. ..
The probability mass funct1onf(x) must sat..isfy rh e 1wo cone11lmns
(')I :fi( .\.·) >- 0• (It) L1,r(x) = J ( 12.2.2)
value x.
11,here 1he sum
ma l 10n 1s ta k·en over all possible
n of a
Th e fol low ing table shows the discrete, probability distributio
//111sm11io11 _J.
. .
random vanable X:
7 II Total
3 6
\lnlue (x)
)
0.5 0.2 0.1
Probabili(v (p) 0.2
d die is
If X denotes the 'numb er of points obtained when an unbiase
Jllustratiorz 2.
of Xis:
thrown', the probability distribution
5 6 Total
3 4
l 2
No. of Poiflls (x) 1
1/6 1/6 1/6
116 J/6
Probability (p) 116 I
I
n as
This distribution may also be writte
f (x) = 1/6; . (x = 1, 2, ... , 6) iased
X den ote s the 'nu mb er of heads obtained in 3 tosses of an unb
lllustrarion 3. If

coin', the probability ·distribution


of Xis '
r-- -
Total 3
I 2
No. of Heads (x) 0 I
)/8
3/8
3/8
Probability (p) J/8 12 ·8)
en by the p.m.f. (Example
lliese Probab ilities may also be gjv

. /(x ) = Jc,Gf o.
(x = I. 2. 3)
atedJy until a head appears. If X
ution
1/Ji.sr,ar; 11
An unb iased coin .is thrown rehpc,d' then its probability distrib
dtnotes 1~ ; ·
firSI ea
11f c number of tails precedmg the I
I

~ below: r .... Total ...


No -:~ 2 3
' Ofh11fl (..t) 0 I

l
- (¾)' GJ' (½)' ... (½f ... I

2
--~ •
,., . -~ _1 ':J I: I ca .I /1/H I: J,odfJ
1 1
, .J • l'.r - . '
428
00□
This distribution can be wrilfen hy f.he p.111,I'.

/(.r) = ( ~ r•I (.r = o. ,. 2..... ,~,J


!Note: (i) A discrete random v11ri11h.lc (I'. v.) eun w;,.,,,,,,e dtlwr ~• 'Ht1it,~' IHJ1J11)4ir, f 1.
· I I · 1:• 'I • 1111 I i I he,· , ,/· v I
1
f//1 ' 1 >'1. :• ·
val ll~S ~ sec II lust 1:a_tion~· I, 2' 3) lll' II ' C(~I II IIII ') y 111 I II I ,L. ' ' .' • ' • ",' It!:, , 11~-~(ll//11111/ :·
•j

-~?:
a clelm1te probah11lly (1.c. a non-ncga11ve rcul 11111t1ht.1) .11-;~ouurcd w11l1 1:,1d11IJ1;J,j,.11_J, 111
i,,,,.
the variabk. such that lhc tollll prohahilily Is I. The 'pn.,l~:·1hilif y di:;trih1,ti,111' ,;f:, ,li:_/' _ ~: 'I
• shows ho\\' the total probnhility I is distrihulcd over lhc! d1llcren11Jo,;:,ihfo 'IH/11,~is,Jt't/µ·. 1ltJ.t:1,
The probability or an int~rval of values is then gi vcn hy the tmm of tlw pr"h:,l>ilit'j ,,~~/'.J•..I .~ ~n~:/4.
with those \'alucs which lie in 1hnt interval. "~·,1
It will be shown later that n conti1111011s rundrnn variuhJe u~i;umcd
. I I · 1· . · . I I ' I .
m,
11,u:,mm:,l;J,1 ,. . .
· lll/111,:'i
num b er o f possible values. Pro rn ll I1Ics arc now w.;:-.oclil cl Wlf J 1111.crvuh ,,J' •i:,Ju,~ 4• ,- · ,

with each individual value ol' the variithlt!. The prohahilily i:; 11hl,111wtl hy in,,•, ·:' .'!!" f.ll
I b I·1,ry
' pnna I
· ll'IISll_\ ·
· ' .fllll('IIOII · over I IJC 1llllll
· .<I.I.)
· · (p ·
• . li O f' I.IIC llllCJ'V/1, I ,,i!J,,lmz
. - (,i:: i• •

(ii) Generally. the capital feller X is used 10 dcnolc the · rondom wtrlll/Jfo' :uµJ ,111•• ,..,
• •
Jetter x to denote 'mn· spt•,·ified l'f1l11,· ' of 1hc rnndom vuriahlc.:, II' howev,)r ~ · . ~,,m
no t.'t>lll·,, ..· •:m-: ,; ,,
~~,{..

the smaJI Jetter x is often used for both. I '··


Example 12.1 Xis a discrete ra11do111 variah~ havi111-: proha/Jllity 11111.,.,.,rw,,.11,
. ,,n.. \'l'

·--·o- - ... . -- - ·-•


(} I 2 .'I 4 5 IJ 7 '

1
X
-- - -.2k 7A; ·: ;;
1__ _ ..
P = (X = x ) 0 k 2k 2k 3k
--
"2 - - -· . - - - -,
- - - ...... .i. ~u t ·
"-J ion s-Bi n .
o.11.1 a 1
Pois son ,
Normal 429
~ EXPECTATIONS-MEAN GOO
~ (DISCRETE DISTRIB:u......A:ION) VARIANCE, MOMENTS

. discrete random variable x assume ti


Ld J • . . 1e values r r . .
! ,, . .... p,, respectively. Then the expeua( : 1• , : · ·· · • .\ ,, w11h prnhabilitie,
r1· - .. <l h w11. or etpeu ,J I . . \ - Wnth:n
£(.\·)-is detme basb·1· t e sum of product _ . .·.. <( ra 11e o l
s o 1 t 11c J1tl cr~ nt
corre:·ponding
,
.
pro a I ttles. t.: va I UC\ nl . x anJ th e
E(x) = I:p .r
2
The expected value of .r is similarly' defined . . . . ' 12 ..l 1)
d S 11 h.; sum ol prnLlucts r th
of \'alues amJ the con-esponding probabilities. · · u e s4u..irc\
2
E(x ) = L/J.\'.
r I
2

In general, the expected value of any function g(.\ 1 is defined as, ( 12.3 .2)
E[g(x)] = LJJif~(x)
Hence. the expected value of a const~1nt k is the constant k itself.
E(k) = k,
where k is a constant: because E(k) = "E.p;k = k 'f.p; = k.
If the p.m.f. f(x) is given, then the expectations are defined as follow s:
E(x) = rx. f(x) ( 12.lJ)
E(x2 ) = u 2 . f(x ) ( 12 J.4)
Mean of a probability distribution is the expected value of .r.
Mean (JL) = E(x) , 12.>.5 )
Variance is the expected value of (x - µ ) 2• whereµ is the mean.
Vaiiance (a 2) = E(x - µ)2
It may be \hown that
a l = E(x2) - µ2 1 l2 J .6)

The positive square-root of variance gi ves the standurd cltTi([{ion (er).


Momems of a di sc rete distribution are defined as follows :
r-th moment about A: µ;
= X(x - A)"= 'f.(x - AY f( r)
r-th raw moment: µ;. = E(xr) = LX f(x) 1

r-th central moment: µr = E(x - µ)" = l(x ·- J,0,. · .J(.r)


whereµ=: E(x) is the mean of the distribution . Note that
µo
= µo = I: µ; = E(x) .= µ
µ1 = E(x -p) = L(.r --µ) . /(x )=~r . ( ,d - jt~ f t\ )
=JL - µ=O · I .
. no tH.L·ntnl
. Th c central momen ts can be obtained trom '
nwiih.'nt:-- u, m~ the fl'

JlllHh
l)ee 7.2.2)
; <~
µ2 = µ2 - J11 ,;
µ = µ ~- Ju~) l ; + 2Jt 1·
\ l ' - ' , I

· ·/ µ/ µ / + 6u~p;- - 3µ1
µ = µ4 - 4 -~ I ' - .

EXatnpJe 12.·>, ➔ ·, ., i •f,. , '·,-u\1t1 ,J1 ' ·


I ,a h't
/'·11/oll'ill~· ;,ro

f.'i,1cJ 1J,
-~-ili
-- --
,, random ,,, triahle lw~· tiI e

Pr,,lic1hilir ,,
-- - -
0. I
--
-
-
0.3
5
-_--hie
, ·, "' oj the ,and , rarw ' . . , ·uI
f,

0 ..J
-- - - · --

~- s {)

,,,,
,
- ·,

,e e.rpt'ctatimr and the sr, mdard dni. lltc lC.Ll .. B.Com. (Hons} ~
430 Sta ti sti cal Met hod s
El□ fil
I)
Solution (Note that rhl· sum of the probabilities is I; l./J; i::

Mean (JI) = E(x) = I.p; x1


= (0. J X 4) + (OJ + (0.4 X 6) + <0.2 )( 8)
X 5)
= 0.4 + 1.5 + 2.4 + 1.6 = 5.9
2
E(x 2) = Ip; x/ = (0. 1 X 4 ) + (0.3 X f!) + (0.4 x 62, + 2
= 1.6 + 7.5 + 14.4 + 12.8 = 36.3 (0, :q i)

-:-- - ~ c~ · ) = f(.,1 )-W , u~ing (l2J .6)


= _C,.3 - (5.9)1 = J(d - 34 .81 = 1.49

S.D. (a)= .)1,49 = 1.22 Ans. 5.9, I~

/L ~ For "'""' l'llllle of t1 ll'ill th1· f1111ctio11 l!l:l


\. Wmpu ·, 12.3
1
f(x) = ,Lr; .r = 1, 2, 3, ... , 11 be the probabi/i ,
f1111c1io11 of a discrete random mriabl<' x? Find th<!
mean and variance of x. ry lllaii
[W.B.H.S. '78, •g~
. arc
Solution The conditions for uny function/(x) to be a p.m.f
(i) /(.r) ~ 0, (ii) I/(x) = I
x is negative.
From (i), we have a > 0, because none of the values of

From (ii), J= 1:.f(x) = !


-'=I
,v; = a Ix= a. n(n + ))
I 2
2
a=
11(11 + I)

f~
"
Mean =E(x) =u f(x) =
t,
L x . ax =a .r2 L
,t =I I

= a(J 2 + 22 + ;.. + ,,2)


= 2 . 11(n+J)(211+l) _ ~ -t f)
n(n + 1) 6 - Y ~
n

Again,
2
E(x ) = Lr2 ./(x) = a Lx3
I

= a( 13 + 23 + ... + n3)
+ 1)} =
2
= 2 11(11 11(11 + I)
1l II + 1) 2 2

Variance = E(xJ2 - IE(x) 7=11< 2+1) _ ( 2113+1)'11

(11 - I) (ll + 2)
= 18
01
11•s:
Example 12 ·4 A random variable X is defined as _1<,l/0
i f111
· Prob (X = 1) = p, Prob (X = 0) = 1 - p
µ µ;, µ,
where Q < p < 1 F:'Ind the mean, variance and the cem ral moments 1'
d' 'b ·. · ' '
. I_Slrl Ult on.
eti ca l Di str ibu tio ns -B i
1'l1eo r: .
nom .1a 1.
Po iss on,
Nor mal 431
roba bili ty dist ribu tion of X is as f
fhC P
~~pof1 Value (x)
O11 OWs: GJOIII
1 ---. --- -- .
0
pro bab ility p Tot al
1-p
rnornents are
~ ~\\1 Jl; = E(X ) = p X l + (1 _ p) X O _
, E(X 2) - P + 0 == P
µ2 = =p X l 2 + (l - p) X 02 - .
3
µ'3 = E(X ) = p x 13 + ( l -P Xl +( l-) 0
- p) X Q3 = p /J x = p
µ~ ~ E(X4) = p x 14 + ( l - p) X Q4 = p
Using (12.3.8)
µ2 = p - p2 = p(l - p)
µ3 = p - 3p. p
+ 2p3 = p - 3p2 + 2p3 - (1
- 2 4 .
p - p . p + 6p . p - 3p4 = p( I -- p)p - p) (1 - 2p)
µ4 - (1 - 3p + 3 . 2)
Also, Me an( µ) = E(x ) = µ = p
1 -. P
2
variance (<1 ) = P-2 = p(l - JJ).

Example 12.5 [!][ ]ii)


Ob tain the exp ect atio n of the num ber of
head in an indefinite ser ies <?f tos ses of the tails preceding the first
same coin. · · ·
[C.U., B.Sc.(Econ) '81 (New)]
Solution Let the ran dom var iab le X den ote the
num ber of tails preceding the first head.
Suppose that p denotes the pro bab ility of gett ing a hea d in one toss of the coin and
~notes the probability of a tail . The n the pro q (= l - p)
bability that x tails precede the first head is
P(X = x) = Pro bab ility of x successive
tails, followed by a head.
= q . q . q .~ .. , q . p (q's nothing x times)
= qt. p
The p.m.f of Xis thu s giv en by
f(x) = ptf ; (x= 0, 1, 2, 3, •·· oc)
..
[Check: (i) f(x ) ~ O; bec aus e none of pand q is
· nega t1ve

(ii) L f (x) = pqo + pq• + pq2 + pq


~
3
+ ......
0

= p( 1 + q + q2 + q3 + ..... .
1'ric . = (1 _ qrl = p. p- 1 = 1]_ .
C):Pt!ctation of the n~m berp of tail s pre ced
ing the first bead is
00 .

E(x ) = :Ex . f(x) = L/X ·pq'C


0 ..
= 0 +· pq + 2pq2 + 3pq2 + ... ...
= pq( l + 2q + 3q2 + ...... )
. 2
= pq( l - qt-.., == pq . p - ::: qp-1
Ans. q/p
= ·qlp .
432
Sta tist ica l Met hod s
EDII
® UNIFORM DISTRIBUTION (DISCRETE)
sx
If a discrete random variable x assumes n possible value
r~~~· :··, x,. with
'!(ohabi/ities, then the probability that it takes any particula
he probability distribution defin-ed by. the. probability mass funct is a consia~~
- . ion (p,fll •),
f ,,,
.
f( X)- Jin, (,\-X 1,X2 ,···•x ,,)
(l2-4.
is known as Uniform Di.uriblllion. 11

Tab le 12. 3 Uni form Dis trib utio n (Dis cret e)


X x,, Total ·-
1/n l/11 Jin -
fix)
unbiased die follo:su .
. T~c n_umber of points obtained in a single throw of an nrf~
d1stnbut1on with p.m.f
f(x) = 1/6, (x = 1, 2, 3, ... , 6).
No. of Poiflts x I 2 3 4 5 6 Tolal

1/6 1/6 1/6 116 1/6 J/6 I


Prob abili ty J(x)

a constant (independent of zhi


[Note: (i) The p.m.f of a discrete uniform distribution is
values of x.
this constant is one divided by the number of possible
s of the random variable maylr
(ii) In a discrete unifom1 distribution, the possible value
have a common difference. HowM-
any real numbers, and the successive values may not
there must be a finite number of possible values.
Example 12.6 If a db;crete random variable xfoll ows unifom1 distributionali
only the valu es 8, 9, 11, 15, 18, 20, find the prob abilities (i) P(i =9i
assumes
(ii) P(x = 12), (iii) P(x < 15), (iv) P(x ~ 15), (v) P(x
> 15), (vi) P(lx - 141 < 15).
values, the probamiiJ
Solution Since x hac; a discrete uniform distribution with 6 possible

the probability that it takes a value~


that it takes any particular value ~s a constant¼ and

than those given , is zero. Therefore,


1
P(x= 9) = -
(i) 6
(ii) P(x= 12) =0

(iii ) Now, x < 15 implies that x


takes only 3 values, viz. S, 9, l l.
1 1
Hence, P(x < 15) = 3 x -6 = - • 8 9,
11.1!-
2 1
• ' 5 · tuded) ' vii '
(iv) Again, x ~ IS implies that the variable takes 4 value s ( 1 me
Hence
I 2
P(x S 15) = 4 x 6 3 =

(v ) x > 15 impli es only ~ possible values of x viz. 18 and 20


I I
P{x > l 5) :;;: 2 x 6 3 =-=

. r •
ore t ica l Di str ibu tio ns- Bi
. nom lal' Poi ss
,;he on, Normal 433
erence 0 f
[!["JI
14 1< 5 implies that lhc diff x and 14 is lo he lcNs th·111 ~- . lake., only
,,·i) 1.r-values I I, 15 and 18. Hence. • • , so ..r
che
P(lx - 141 < 5) = 3 x ! _!
6 - 2

'111.1·. ! O ! 1 I 1
()'
1
2'3 1
J' 2
r;r-,r.a
Je 12, 7 Find the mean {.'•II{.1 I 1le VI I L.:JLJu
= 1, 2, ... ,,) · wu <ml del•lation r,f t',,e u11,1:1;or,n
11rionf(x) = l/11; (x '
solution Mean (µ) = E(x) = l:t . f(x)

= ! X. (1/n) = (1/11) !
"., I
X

= (1/,r)(I +2 + ... +r)


= 11(11 +I II + I
2,, = - 2

£(x2 ) = t.t2 .f(x ) = !, x2. (1/n)


x• I

= (1/11) (12 + 22 + ... + 112)


= 11(11 + I) (211 + 1) - (11 + I) (211 + I)
611 ·- 6
Using (12.3.6),

u' = E(x') _ µ' = (11 + I) ~211 + J) _ ( n ;i )'


2
n -1
=-
12
112 ✓(n2
S.D. (<1) = ✓( - I) I 12 Ans. (n + 1)/2,

~ TIO N
~ BINOMIAL DIS TR IBU
Bina · I tion and is defined by the p.m.f.
nua distribution is a discrete probability distribu ( 12.5 .1)
Where f(x) = nc, pl q1-x,. (x = 0, 1, 2, ... , n)
. ·
Su: and q are positive fraction (p + q = I).
t trails in each of which the probability
of OC c~ that we have a series of 11 independen
C\'ent llence of an event is ·fixed and con
stantly p. Then the probabil11ity that the
re
9-:: I _occurs exactly r times in n
trials (Section I l .8) is xc, p' q - ', whe
es O, l, 2, ..., n. ln general, the occu"rren ce
of the P and r may assume any of the valu ''fi ·1
· known as t11 ure . ,c TI
I
&en-...even tVis •ca led ..success" and its
non-occurrence 1s
"""illl.Scd .. , . . .. , . .. . .
· hi aSC,i anou~ of the den theory is as follows:
I

m euch tri ,11 1s u


c°"-slQ ~s of n mdepen t rrilll.r if the probab1hty of success
q, then the probability of x successes
lit,d 0~'?' and the pro bability of 'f~ilure' is omial distribution (12.5.1.)
VsousJy 11 _ x fa ilures) is give n by the Bin
r
X
I 2 11
f(x) q'' "C, pq" - I pfl
I

of _Note: (i_) In the expression for /(x), besides the factor "C,, the power of
P ('
success) is the number of successes x, ancJ the power of q (i.e., proba
bilit 'r•c., Probabil'11
number.. of failures
. 11 _ ""
•\.,
YO failure). Y ,,~
(~~?The sum of the powers of p and q is always 11, whatever be the number of
sueces.•~.
(~II) These are (n + I) possible values of x, viz. 0, 1, 2, ... , n.
(iv) The total probability for ull the. (n + 1) possible successes
is I.]
b
. Th~ distribution is known as 'binomial', because the probabilities are given Ythe
b1nom1al serie s., .
(q + p)" = q" + 11cl pq"- I + "Czp2q11-2 + ... + /Jn. 02.5
1
of a rando m expe rimen t with a constant proba b/
A serie~ of independent trials
is kno uy
of success 1n each trial is called a "Benwullian series", and this distribution
ulli. Wn
as "Bernoulli's distrieution", after the name of its discoverer James Berno
as
The two constants n and p appearing in the expression for f(x), are known
are known, the
'para ,nete rs' of the binomial distribution. If the values of parameters
distribution is completely known (q = 1 - p). ·
ns:
I should be noted that the binomial distribution holds under the following conditio
(1) The result of any trial can be classified only unde r two categories; e.g.,
either
head or tail in a throw of one coin; either a "six" or a "non-six" in one throw
either a spade or a non-spade when one card is drawn; either a defective
of a die;
or a non-defective when an article .is examined from a lot; i.e., in general,
occurrence (called success) or non-o. ccurrence (called failure) of a certain
.- .
event.
(2) The robability ofsuccess in each trial remains a
constant, and does not change
r. or examp e, the proba 1 1 - btaining a head in
from one tna o an
1 .
successive thrown of a coin is always 2 ; the probability of obtaining adefecttve
article from a batch does not change.in successive drawings with replacement,
and practically remains a constant even in drawings without replacement, when
the batch is large . =, .. .
5
.

~~f
(3) The trials are irJ.depetulent, so that the probability of success in any 1 0
·unaffected by the results of other tnals . For ex~ ple, in.successive ~ bilitY
a coi~ the occurrence of a head at any trial will in no way affect the pro ;e!ltr·
of a head or.a tail in any subsequent trial; or if several coins are thrown t~ter rltt
the occurrence of. a head or a tail in any particular coin does not a
probability of occurrenc e of a head in any othe·r .coin.
ret i cal Distributions- Binomial, Poisson,
rneo Normal 435
_ np /variance= npq /
a.
Menn - • / (12.5.3)
(ll d rd deviation (a) = .J npq
s1an a . . .
. rnial distnbut1on may have either one or two modes Wh ( l) .
(3) B1~01eger Mode is the largest integer contained there1·n. H en n + hp is(not
an 1n ' . . owever, w en n +
1)p is itself an mteger, there are two modes, viz. (n + 1)p and (n + )p _ I.
q-p_ 1
. 1-6
Skewness (r,) = c= , Kurtosis ( y ) = pq
l "npq 2
(4) npq (12.5.4)
th
When p = q = 2' e diSlribution becomes symmetrical for all values of n.

i.e .. J(x) = f(11 - x) = "c,(1J


(S) If x follo~vs bi ~omial distribution with parameters (n , p) and y follows binomial
1
distribution with parameters (11 2, -p), andx and y are statistically independent,
then (x +y) also follows binomial distribution with parameters (n + ni, p).
1
The result can be extended to several 'independent binomial variates with a
common p.
(6) Binomial distribution may be obtained as a ·limiting case of Hypergeometric
distribution. / · ·
Example 12.8 Three coins are tossed. Find t~~ probabilities of (i) 0 head,
l head, 2 heads, 3 heads; (ii) more than one head; (m) at least 1 head.
S I tion Here the 'random experiment' consists in tossing 3 coins and observing ~e number
o~~ds. The 'r~dom variable' xis the number of heads obtained in a tos~ of 3 coms ..
If we denote the occurrence Of head as "success" ' and the coins are assumed to be unbiased,
then
robability of "success'' in a single triaJ
p = P l
= Probability of head w_ith a single coin = 2
I 1
q = l-p=l-2 =2
f . dependent trials = 3 .
.
n = Number ochmcoin and the tna. Is are independent, the variable x
Smee the value of p is constant for ea _ !
Therefore, the probability of
tollows binomial 3
distribution with parame ters n = , P - 2 ·
.t SUCcesses is
l
J-r l
l x 1 .- Jc ( - )
f(x ) = •c,v· q n- , -
. -
3C
, CJ
- ( -
2
) - ., 2

respectively,
(i) Putting the values of :( = 0, 1 2
, I I

. . ·or Osuccess
f (0) = Probab1hty
3 1
. ( 1) 1 x! == -s
= JCo 2 == 8
f 1 success
Probability o
/ (l):::
436 8tt1ti o t.ic:tJl. Nc t l1oda
0011

= J(_,:, ( 2I J' = J X
I
H 3
= 8

/( 2) == Probahility of 2 successes
l):1 I 3
=.JC7. ( 2 = 3 X 8 = 8

/( 3) = Probability of 3 successes
l):1 J I
(
= J(,'3 2 =JX 8= 8
(ii) Prob (more than one success)= /(2) + f('J)
3 I I
=-+-
8 8
=-2
(iii) Prob (at least one success)= I - Prob (0 succc~;s)
I 7
= I - /(OJ= J - 8 = 8

fill
Examp le 12. 9 Five coins are tossed 3200 times. Find the expected frequtr.ci~:
of the distrihution of heads and tails, and tabulate the result. Calculate the rritz.
number of heads and standard deviation. [C.A., May 'i;;
Solution We assume that the coins arc unbiai.;cd. The prohability of getting a head lorriif~

~ for each coin at caCh trial. Also the successive trials arc independent. So, the corulitia.;ct
binomial distribution arc fulfilled. The probabilities of Ohead, I head, ... 5 heads arc then~;o
by the ~ucc.:c:ssivc terms in the expansion of the binomial expression

C~J + = .GI~.sc, ( ~)( ~J + 'C2 ( ~J( ~J


+_~c1 (~)' (~)' + 5C4 (H(~)+[t
__ _ 5 JO 10 5 I
+ --
32 + - +- +- + -
J2 32 32 32 32
The prohahiJitics of the <liffcrcnt numhcrs (>f' lici•ds ·,nd t · 1 · · 1c tu·,, of :'~ co;\1, t.'.
a1 s m a smg
• • •
the c1t pc,led frcqucnc1c~ rn 3200 tosses arc shown whclow:. , ·--
-----.. -
·---
'1'I
. . i r· cl l J) i .r. t :· i 1J u t i o n s- !3 i n o mi a ] ,
r J1C• •
('l ,,.. t.' '- Poisson , Norma1 437
m ■ I I
I
11 t'l(_·r of hc,Hk and !he !-l:mclarJ dcvia1ion arc now calcularcd from th~ following
fl tC
n,c;1 J1 nt1•
I 1.t n h11t1<'I 1
(,cc E,amplc" 5.57 and 6.29):
~ -
,. .
I

nd ''- --- - - -- () - ---: l -·- --1-·- -- - - - -- - - I'


fr~•piC, . , . r_.-if !fr•" I \ (\l - 3 •l 5
,,iJ,r Tora!
:1 . ) 100 )( )() 100() 1000 500 100
1(11(1 (/
1200
[ 1(,/1 ----
/
~ _.-- _ .,- · ·<i • s1andard deviation (a) ==
. ( --: ) -
✓r-,t / 2 == J. J 2
,1.-;1fl .1
rtd
·• , 1h;1t 1hr ,m:ana '- .ti. cakulall'.d from lhc lll'q11cm:y distrihution ngrcc c., actly wi1h
l~olr, . . (l:!.5J). " 11 ·

~ f,~uul,oc "I' ~ ~ 5x : 2.5, s.d. = J,, ,, = ~5 x .!_ x .!_ = fJ!.


~
11
2 2 Y2
l•B ■
11 ~.10 Th,· m·cra/1 p,·rct'lllagc of j,1il11rt•.\· in a certain cxaminatimt i.'f
//(/ f ,\ till' 11rnl111/,i/iry that ""' ,f" 1,:m11p ,fr, t·,mrliclmt'.\' ar lt•,1.\·J 4 pa.'i.rcd the
o-nrui11 atimr.i;? ll.C. W./\ . Dec. ' 74]
SLJ/utioll I.ct \I!'> dcnoh.· t_h~ ur a l',rndidatc passing. lhc cxa111i11atio11 us "succcs.s". lt i-.
l"\'t:_nt
rrquirL-.1 l'1 !inti the pruhah1hty ol at h:ast 4 sucn•sscs.
11 = Pnihahilily of failurl~ of a canc.lidatc
~ .rn,:-, = .iw I no = '2.15
I' = l'rohahility nf sun:css in a sin!,!IC lrial
= I - 11 = I · · 2/S J/'5 =
11 = Nu111lx·r of l':mdid:th!!'> in the group = 6

ll~m~ hinominl distribution. thl: prohahility nf .,(. :ll)l~l"c(s:c)s''~~ C(' f_r,,')~Q:'f ~


.re,) = ,1c, ,,, ,t . , = ,.(.·, i ~ ~ ~'"5'. (37.,,
.. . . . . . • . .• ·,her 4 • or 5, or(,'. the probahili1y of :.it lc~'\t ➔
Sto,·l· ':tt l~asl •I III a i•roup nf 6 •m.pl~c_s_ -~..' ("1) + j'(S) + ,f((i). Putting x = 4, 5, 6 in the
1.#:n·~\l'~ 1s ~1\'cn hy the sum of proh.1b1l111c .f .
njl:':~~11111 fllrjtx). WC have 4ur.,o

\ ){4) = ''CJ ( H(H •


= 15 )(
34 X 22
5''
en
= 15625
I

.b
·✓
•·c:, (Hrn
' .
3' 2 ( ,' 1

~ £-6) =
)' X 29~ A ( '
'--T '-4' ... . :
= <, " 5•• = , 5625
V
--!)"> = •·c.. ( ~ r(~ J II

=Ix T
J'' X I 72')
= I 5625

1701
J l2 5
0f ]□
'\. <11t )
·. . p 1:' 12.1 1 /11
.
/ {J iwlt•/1t'tuln1t thro11·., (!
. . r
/ ... . tf, ,·
ti rfr/ 1'< /1\'1 ' t I, t • •
l' ru /-,,1!,ili1y
-, /,i/in· t!, ur u,1 i'l' t ·:: m an,1.}1
I .

r. !~. · .
' '·· ,, ,. t1: ,,qJ,,•.,. ·,,·ill " / " '' ·,u
-
·1 r1111t·,· , ._ r,, .·,, ,'' tilt ' / 11/J 1(l/, .,.
, , . ,
iri // ,,, ,r ctf)/1 1'111 . ul. . ' .' ' 'I
li t • •
I ' f , 11 111 1111 •
. . ::,•o :·
• · •. I , , , : i n ;!• , . r;,," 11i, •, ,,,·,Ii""""-·· r/;i1 '
1· 1·1•11

. .,:,_.Ill ,1,~1 ,w•. (ir rli.· "" ,·.


438
l•I ■
Solution Let lhe occurrence of an e,·c!n number be cilled :i - - ~ -. ,,~~• ~-=:t
\ ~? ~- .
proh:ibiliry of fening 3.Il e\'en num ~ (ie_. 5-U- .~
p =
q = probability of not gening :?.Il ~,·en numbe:-1
i.c_ fa:1::::, ~~
11 = number of throws = JO
:~s~ ~, 10 tr..v-.,> :.)
Using binomial distribution. the probability of .r 5-u
j{.t) = wet pf q10- r
It is given that
/(5) = ~ x/(.i )
10csP~ .. _ , x 11,c p' qt,
or, 'T- - ' ,
10! ~ 10! , 1,
or, 5 ! 5 ! p- <r = :? X .i ! 6 ! p q·

Simplifying. we get 3p = 5q
or. 3p = 5(1 -p)
:; 5 1

Solving we get p = t .so lhJl q = I - P = 8= S I-


all (i.e. 0 s-.:cce» J :.i
Probability that an e\'en number will not appcJI Jl
( .,)10
/(0) = ql[l = lt
de1·iarion of bir.or.i.icl ~~
Example 12.1 2 Find the meall alld the .standard
with parameras 11 and p.
Solution The p.m.f. of binomial clistribution is
j(.T) = r.c_ tfrq" -.r;( .r:0, 1,2, ... ,n)
Hence. by definition (12.3.3),

Mean = E(x) = L x j(x) = !


I r: 0
.T. "C.rp1 <f-.r

= 0 x ("C0 p q") + I x ("C1 p q1- I+ 2 x ("C~ Jr


0 q2 -=) +
1

3 q" -.l) + ... + n X ("C11 p" q0)


3 X cncJ p
., n(n -l)( n-2 ) ~ ~ i
_1 n(n -1) ., p-q"'-- ... _-r.1_
+2x lx2 p-q "--+ 3x lx1 x3
=O +Jx npq "
., n(n - l)(n - .,,
= npq"- + n(n - I) p2 q"- - +
1 ., - p-' q"- ~.;. ... + n p'
1x_
, n(n -l)t n-2 )
= 11p{q " - I + (11 - I) pq" - - + I X2 p~ q" -3 + ... + p~ - I )

= 1tp [q" - I + 11 - I Cl p~ - ~ + n - I c2 pq" - .1 + ... + !:


_ I Cr. _ p" - I J
1
= np(q + p)" - 1
• repla cing n by n - I in ( 12.5. 2)
= np ( I)'' - 1, because p + q = I
= 11p(I J -= np
Therefore, Mean (JO= np
Again , oy ( 12.3. 6) a ~= F:h .-~) -· µ !
£ ( r ) -::. Er ./ (c) .: t j.t(.r - 1} + ..c) ./(-'
)
Bur,
-: L l{.l ·- l) / (.t) + !.r. / ( t)
:: 1:.t (X - l) , j ( t ) ·l µ.
;he ore tica l Dis trib utio ns -Bi no~ ~al,
P.:,i!3son,
Hcr.:.a.1 43 9
n 1•1 ■
. . <"r(J -1) . f(x) = L.r (x - 1).l'!C
~~- ~ nlr/'o - r .
I
_r:Q iP ~ ,
:: O(- l){' :Co l'cf }+l x0{ 11 C pq" - 1,
I I+ 2 X 1 l"C1 p1 q" - 21
+3X 2("C pJ,.... . J,
JPl(n - }) } } •t + ... + ll (n - }){ 11 C p'I </'}
= 0 + 0 + '2 x 1 l
l ,._.,
} X 2 /> C/ • + 3 X 2 ~ - l )(,a -
\ (
2) "
i 11-~}
+ .. . + "' 11 - 1ll I . I'"\ 1 x 2 ,: 3 q
= n( n - 1> />"l <f., - : + ri(,1 - 1)( -, 1
= ,i(,1 __ \) />i l:.,i .. 2 + ( n - .) P' ,( - ' + ... + 11(n - I) /I'
., II - 2)pq" - ~+ ..ri - :
= ,,,11 - 1) />: [q" • i + n - =c n11 - ·• · + ,, I
.
= 11(n - l)p. :(q+ )" - 1 ,r., . + ... +" -?c,-21 -~
3

- = "~ .. repl acin gnh yn-} in(l2 .5.2 "' l


- u( n - 1) p ( I l -•. smcc /> + 'I == 1 )
= rr(n - 1) p1
£L,f = 11 <11 - I) /J"l + ,, .•md h~ncc
a1 = E(.\~ > - ,,1
= 11(11 - I) /'!+ JI - µl = 11(11 _ I) /J1 + np _ (np)2
., "
= rqr - ~
llf'" + 11/1 - ll"lf'"l : - II/>~ + tlf'
= 111 1( 1 - p) = 11pq. since q = \ _ ,. 1
Standaru Deviation ( aJ = ,I"i''I

A11s. ffil!ln = np. s.d.= ~

\•1 ■
Example 12.13 Arithmetic mew, a11cl standanl deriatio11
of a bi11omial disrriburi on
al? rrspectfrdy .J c111tl .JB / 3 . Find t/,e mlues of q and
p. (W.B .H.S. •8 l l
Solution Using 1h~ formulat! tor mean and s.d. of the
binnm ial distribution
np ; 4. ..J11pq = .Js / 3
. Squaring the ~~(on<l n:1.uion. rrpq = 8/3. Now pullin
g 1111 = 4 and solving we h;m! q = '!.13.
H~1-
-1.i·.,: , p= l - ..
,,=I I.,.
Puttin g the valu\! nf /> in "I'= 4, we gel ,1 = 12.
t\11s. ,1 =- I:! . P = 10.

i L!L■
find th,! mode of hi,romial tfotrib11rio11 ,rif_l1 />.l/'£1111eras

I
I

i t r _, i)
~,:
' ' :,
-- .
, ti

l
_) \ \ / 0-,
l_ I . ' ,'
440 Sta t i s ~i cal Mewicd s
[. I IJ
n! n '.
r
= r-! . 1{ . r-1
(x-1)'. {n-.c + l ) !P .r! (n - .rJ ! Pq" ·r

:cq :rq
= (n - x + l)p = ( ;7 + l)p- .lp
We fir.d that
(a) /u - I) < ft.ti
,, henev.:r xq < (n + I l p - :rp
or. :r(p + q) < (n + I) p: i.e. x < (n + I) P
(b) ju: - I) =J(x)
whenever .rq = ln + I) p - :cp: i.e. .r = (n + I) P ·- II
(c) j (.r - I l > /(:r)
whenever xq>(n +l ) p-.rp: i.e. x>(n +l)p

We di.."us.s the follo\\ing ..:as.es:


Cc.se I U7:en , n + I )p 'is :wr an inugtr.
Let r be the brgesr integer in ,·n ~ I )p. Since in a binomial distribution . .r assumts cc.j
integral values. the result t.3 , is impossibk
From (1 J.
/tr - I ) < f(xl. when .r = I. 2. 3 ..... r
or. f(OI < f(l 1. j( I) </(2.1. ... ffr - I) </fr)
or. j(Ol < j<i} </11) < ... f(r- I) </(r)
From (~J-
f (z - I) > ftr). when .r = r + 1. r + 2. ..., 11
or. f (r ) <J(r+ I). f(r+ l)>f(r + 2)• ... f(n- D>j(n)
or. f(r) >!fr+ 1) > j(r + 2) > ... > /(n)
Combin ing chl resuJli ar 151 and (6) .
/iCJ <f(I) < ... </(r- I) </(r) > /(r .... I)> ... >/1_1:).

We find lh:i1 ~ p:-c,b2l:iiliry/<n corresi;<lnding to.r = r is the maximum. and th~i.1c:~


mode is r.
Tons. when fn + I)p is noc an integer, the mode is the largest integer conoinal in lr. + l
r
Cau JI Wht11 f., + 1)pis an inreger; say (n + /)p = r.
From (2.) .
j tx - IJ <j(.rJ. \\hen.r = I. 2. 3..... r-1
or, f fOJ < f{I) < ... <Jfr- I ).
From '3J.
/ L'r - I J = j(.fJ. when .r =,
er. f(r - I) = j(r,
f-rom (..:J.
/fa - l J > f 1...c), wb:-n .r == r -~ I. r -;. 2. .... II
°'• .. / (r) > j ( r ~ JI > ... > f (11}
Com!r.r.m~ lhe r~~-ult.s :!.I (7). 13 ) and (9 J.
f<OJ <fi l 1< ... <J1r - li ·= j fr )> f (r + l )> ... > ( wt ,.,_-.,

~ ~ rm;1 lk.I tht- ;i1c.~ ul.,e\ r .-..,.,. . . 1 -
,,. :-~'I. · ,<-
. . ~>-
I , ' <.j,.JJ Ir - 11<."(l'ff~ po!.'<iirn.! Ill .l' . . ,. .lfd .t =- I~ -.~ • J!jt'\
, . ,. .. , r• .
,
.. - etrJJ .J..:.d h:n c: llk- 13-.r,.,.-s; \ ;,>JL' . t. •
~ . 1 ~e ....u · no\\ l\\ o mndt"S r ..: ,- - : . JJ1 ...
. -· · u••
. •,-. ···-'- r ,
L• •lmlJUuJ •u·.. ha\c r.-u ~ ,
r

1!1[■

( 12.6.1)

\ \ \
<"= \+-+ - -
\ ! :! ! + J ! + ... = '2.71~ approx (12.<,.2)
~ ""'°st~mt r.t appc:-:nin~ in t l ' t\ 1) . , k
.i:;.,."1..~!?\.,.n, \t m;w ix· n"\t ... ,.. ti , l -· . . ts nown :1s 'pnrnm~tcr' or the Poisson
I=\\ l. ~ .~ ..... ·"' with pn~\hilitie~ '\"
• \:, · l,\t l ,~ \':\n·thh.' ·1, ·u , · · r ·
:h ·..
sts ,mes 1\1\ II\ imtc number of valu\!s
-~ • ., S 0\\ ll lC ow:

Table ll. 5 Poisso n Distri bution


\" 0 "I
J Totul

,, . ,,,-
-t!I '
t•· ·m. 111-'
j
.f\\' e··..,. <'~ . 111 ---
.:! ! 3!
118 Jistril°luti"'n is ll3!l\~d afll'r its dis,~"wcrcr S.D. Poisson.
!lipo_ ~t Prope rties
/i l\~~~'\1 distrihution is a discrete pn,bability distribution when! the rnnuom
V ''3:ru.lile ;l$.$lll\les 3 counrnhly infinite numh~r nf values 0, l, 2, .. ., The

OQ .

,.~ :'.li;orfon i~ l~,mple rely ~ix-.:i tied. when the pnramctcr 111 (positi vc) is known .
\/ -:1...:lll = m. \ an anee = m
"' .. St:mJJ.1'1 Je,·iation ( en = Jm 12.6.J) (
-' 1 ~ i~~on di:-tribution may hun: either one or two mod~s (like the hinomh1l
~ ,!rih:.tri(,n). \Vhe m is· not an integc.!r. mod~ is the largl~st integer contain\!<l
11
i., r.i . Howe\· ~. when ,,, i~ itself an intcgl!r, there nrc two
mmks. viz m and
~ - l.
.{, Sk~\\ll~ . ._ (v ) L"' • ("' ) _ _!_ (1'2 .6.4)
= r . f'.ll l10S1S 11 - 111
'· 11
vm •
·' • /4h~;._ Poisson distribution is positively skew and lcptokurllc.
'-/ r Jand _ ,-
~ \' ti\'.a\,· th
=
inJ,•pin,knt Pllisson varinhk s with paramctc'.' "'1 and ,"'.'
) l , ,,,. p · ~..: l)ll distnhu twn with p:mum:tcr
,,.. " . c!tl lX + \' :\ so ,o110 ~I 01 ., ., •
;: ._ ,, ~ m ) .
" -. ·
1
~ ~-;~tt ihuLi on may he u, ~d a>:m a~P" ~, ;mat ion to binomial <I istributiun.
1 •~ ~m:.H. ~md 83 ,~ llr-:;;1.· . hul 11]> tllltl, ·
Sta ti s~ i cal Net.hods

Some examples of Poisson \·ariable are:


I iJ \"~mber of printing mistakes per page (or typographic al error~
ski11ed typi st): per Palet>i .
• •
(ii ) Number of goals scored in a fool ball match:
.
(iii) Number of telephone call~ rccci\'cd in a telephone box per unit inttr.~
. lf
time during a husy period :
(iY) ~umb~r of suicid~s (or death_s fron~ n rar~ d~sc;~~) pe_r year i~ a gi,,en r~.
(\") Num ~r of bactcna pn:sent m a g1rcn hl)u,cl cuhun.: per umr arcJ ob·.;,er,.,·
under the microscope: ...
(Yi) Number of radio-actiw atoms decaying in a gi\'cn interval of time;
(\'ii) Num ~r of defects per unit area of sheet nwtcrial (c.r.. paper. cloth. or~
sheet):
(vi ii) Number vf cars pJssing thrnu[!h a road crossing per unit time inren·i!
(e.g. 1 minute) during a husy paiotl.
,fhple 12.15 :\ mm/n m mriabh• xfollon·s Poi!isn11 disrrih11rio11 wirhpara,,,.t1tr

4 ind thr probabilitit•s that the• mr(ab lc as.mmes the 1 lll11es (i) 0, I, 2.
n 3: (iii) e11 lcasr 2. (Gin•n tlim ,,-J = .0-198).
Solution
1 3; (ii) Itri

For Lhc Poi5-son distribution with par.imctcr m = ]. 1hc prohabiliry of x sucm~~


_,.. I
·
t' • Ill
f (.r) = x! x!
(i) Puning x = 0, I, 2. 3 succcssi\'cly. Lhc required probabilities arc
-.' .,o -, I n
/(0) = -
(' . .J
-,-
C .

I
3
=--
= e- (since O ! I .u1d 3 =I) =
0.
= .0498
e-:,_ 31 c-3 • 3
3
/(1) = - !- = - - =3. c·-
1 1
= 3 X .0498 : . 1494
e-1. f c,-.1. 9
/(2) = 2!=- -
2
4.5 X <'-J =
= 4.5 X .0498 = .2241
l'-3. )3 -J ?7
e .-
/(3) = -3-,- = -6- = 4.5 X c· J
.
= 4.5 X .0498 :: .2241
3 stlcce~
(ii) ~in~e 'less than 3· implies 'ci1hcr 0, or I, or 2· , the probability of less than
is given by the sum of lhc probabilities

I
,l
f (OJ + /(1) +./(2) = .0498 + . J494 + .2241 = 0.4233
· . t I j:JSi •
/iii) Sinre '· 1 I ·151 3· · c<> rhe probability ' :i , SJ
1
~ a ~ impht" s 'e11hc:.·1 .2, or l or 4
/ ·;;r'
\~cc~~th~es is given by lhc SUlil of the prohdbiliti;s tl1c.: infinite nu1ntx- . I i t' .
r ()( ,cnn• .
r.mcc e ,um ot the pro b;.i h·1· .
I lflC: ~ for all valuc:.1, of r. \'i1. . 0. I. 2 .. .... CID 1' ' ·
,/ /1.0) i { ( I) +J ('!.J i .. = l
htn~t>
fr:.; ➔ t (3) -t f H ) ~ . ..; I -.no, .
rn)
1 · · .tl4GK . l 4lJ-J :-: .SOOS
J:;:-:-...3: ,,3
t•f ■

pmbahility of no \'.all during the


(I) period,_,.-: .. t::
.. -:.! ,....
o, 1~
c; •. - ·.
/ ( (Jj = ---- =
... • ·-• 1 C - Y 1

(Ii) Probability of cx act l) ..! c.sll~


du:-ir,f th.: r<"noJ (i..r _ t.:: ..:
, l\
(. : · , ~51' DS~l Y: 9.' ~2 5
J<·H = = ----- = .1:?-S

A~1. .<.>~-:: !. . 1.3 .:-6


I•! i9
Example 12 .1 7 IA 'I ,l /,c 'io 1 r,'h~Ute,_J rn
· tlu Pn ino n Jorn:. fj Pl:r = 1 1
: r (.t = 2). wl, at il /'( .\ =
(} or I ). .-\ ho }In tl/-.( t).
Solution For 11oi,,on di' -tr ibu tin th::
n.
...p.m' .f. i\

I I tJ: : --
('

J '.
111

. . . b ·fl l s.e .. :.r,:


. I •1·• .. ~......
The probJhility tk1t \ a,, um c, 1t • , . lu-- I t" th::n g.t' en : J,
,. ~!lo : c, ,...r 1 •• , _ _.
)~- ';.1 - ·
1 =l.

/'(.' l :: JI: : /f l I
· • I
,· nl
- :: e·"'•rn
I '.
· r \ ;1-;,urn~, the ,·atuc 2 i~ / (2) .
, l ..
prl) t1~0
..l-.·1
1 ·1ty ti1.1
t' .
Jl.'ll·l.Jr y, th~
..,
/'u := ~) ;;; ;/f _l
... .. ,,(
(,
L' . ,n· = -
2

rt I j : ,> I --
. ')
,. [!£111
444 Sta ti s ti co 1 J,fo t hvd11
0(]11
Example 12 .10 Find the mcm1, and i•arimu:r. 0 ll'ol.u m1 t1J,1,i1Juu
rc.u., H.Sc.(fa:on.J' ~J· l .f (• ,,,,
, i'YI ·'--''TTI '7

Solution Pois(,on distribution ic. defined hy lhc proh,1'1Hiry m:tv, fur1tri,m 'JJ.rn.;.J S. ·'1-
,,,,
f (x) = e-"' - , (.x = (), I, 2, .. , '/•J
x .
1/,

Mean = Hr> = _L x ·(( :r. J


,-,.,(J

= O./<OJ+ I . ft IJ + 2 .f(2J + 'i • f (J J + .,. "' infinity.


'J
m m m'
· m - 2 ,, m - - + 1 l' "' -1 -! + ...
= 0+ l I'. + ·"
,C 2! . . ,

mi mi
. m + r .. m + tf 2 ! + ...
111
m - . -
=c 2! ,
m m
-,n e- '" ( I +- +-+
1
!
...)
- ., I! 2
= m . e ·"' . r."'
=m
Variance = £(x2) - { I.JxJ J2
= m2, \inc;c /:,'(x) -= m
E(r) -
= Ef.xfa- lJ+..rJ - 111 2
= Efx(x- l)J +m-m2

= El .r.(.r - I JJ :.: f .xf .x - I J • J (.x J

= 0 ./(0 ) + I x<J.fflJ+2 ✓~ I .J(2J+ 3z2 ./OJ+ ...


= 2 . f(2) + 3 Y. 2 . j('!, J + 4 Y 3 . f( 4J + ...
mi m·• rn 1
= 2 . l,-n, -2 ! + 3 Y. 2 . c:- hl - + 4 1/.. .'.\ . t' _,,, -4 ! + ...
J!
m' m◄
= e-"' . 1112 + e.. ,,, . ~ + l' m -
I! . 2 ! + ...

= m1. ,..,., ( , ,. !~ + !
l! 2
!!C·•... j
-- fn: {' . , = m· 1•1 / II 'I

V .
i!mmcc = m• + m __11 / = ,,,
~

Example 12.19 Fill(/ lht mod,-of /'"i\' 'w11 (1·. ·1 .


J,\/1 I l{//itJ '!.
, , .I l 1~(,.. i]
So!u!ion t,11 ·c v · . IC.( l., fl.S r . Wi:1111 ). 1'
,I\'
' r, I I •· ,,1,vm tl 1,1r d111111 ,11 v,11'1I Prll. i llllU ('I ti11· ' 11 i.: ,. J , \ ~ri 1·-., •'
\•.iwt~ I 1111 . "' · l ,IIHl, 111, \' ill l,U
· ' · · · ' • ,md 1/u: pr , ,11
, ,1, l l: ', ,ll l' / '. i\'L'II l1y

111
111 1 ; : h X ~- ., 1 ,.: , . ... _
rl ' • I ' (I t1 . I , .' , . , · ·· l
,,
' / , I,' l h ,, , ,: 1• ' / 1 I •I {11 \ , :, . .
. " l,1 Il l 1
;.'rd ·, ! tll :fl c, fl • ti . .. '1 ' ~\ J 11 1111 1111• , 1
. .
, . i . a m11 1. 111 1; n11 .
rheoret ica J Di stributi ons-B. .
inom1a1, Poi s son
' Norma l 445
(D •
m1-1

f(x -1) (x - I)!


f (x) = e--1".
.-x!
I
e-"' Ill (x~o,
X
,,, ... (J)
We find that:
(a) / (x - I) < f (x). whenever x < 111
.... (2)
(h) f(x - I) = /(x). whenever x = 111
.... (3)
(c) f (x - 1) > .f(x). wlwncver x > m
.... (4)

Cast/ When m is not an integer,


· • d'1s 1n'buuon
Lei r be the intcgml part of. m.. Since in a Poisson .
· x assumes only non-negative
I I
in:egral va Iues, t 1c rcsu t (3) 1s 1mpossiblc here. ·
From (2). f (x - I) < /(x), when x = I, 2, 3, ... , r
/(0) < /(I) </(2) < ... <J(r) ... (5)
From (4), f(x- I)> /(x), whcnx= r+ I, r+ 2, .. ....
> f (r + I)> /(r + 2) > ......
J(r) ... (6)
Combining the rcsulls (5) and (6), we have
f(O) </(I)< ... <J(r- 1) <J(r) > f(r+ 1) > ...
Thus, the mode of Poisson distribution is r, i.e., the largest integer conrained in the
parameter m.
Case II When m is an integer, say m = r
From (2), /(x-1) <J(x), whenx= 1, 2, 3, ... , (r -- 1)
... (7)
j(O) <J(l) </(2} < ... <J(r- 1)
From (3), J(x _ I) = J(x), when x = r
... (8)
f(r- 1) =f(r)
From (4), j(x - I) > / (x), when x = r + 1, r + 2, ...
... (9)
J(r) > J(r + I) > J(r + 2) > ......

Combining the results (7), (8) and (9)


f(O) <J(l)< ... <J(r- l)=f(r)>f(r+ l)> ... ... .
W 1 and the largest of all. Thus the Poisson
. c find th~t two terms f(r) and J(r - I) arc equa . d _1
~•,tnb · 1h modes being m an m ·
Ullon is bimodal when 111 is an integer, e
Not . . . , b'nomial
1 di !;tribution:
e. This method is identical with that ,or (!1['1[]

I lG\ TO BI NOMI AL
I '-)-·! J POI SSON ·APPROXI MATION
, D .
I ~.
iJ:,
ISTRIBUTION
· · . case o
I 8 . mi·at distribuuon
ino
. •1Gn(jj ,. . . cJ ·isj llliil ll llg "
, :2.\ l' :,1J ibu11on ( l 2.(j l )may bcobtame "'
\ (i.J_ 'Hider the follow ing co!tdit ions: . , , -4 cv":
. !fie' 1. i... .. , . • . fi . ·I lari!r : ,., ., ,. ()·
I
_.,
11· )
tht i -
u l0u1: r uf trnl• fl 1•·, ,n 11ut<. Y
. . • ., . . .•
1,, fl· it' I '
cn1.=h srria , . .,
-~ ·

f..., .:l ·
: r: i1 .1 l ,- htit,1lHv oi' ·:; 11 cn·:.•. ,,,) Pi t.: Xtf
0 -
l.li •
~ ·'r! ~an 1,p :-:· m h fl nitr .
446 St a t is t ic ,il /IJct/h)d D

r-r:a
Unde r tliest' n 1nd1 11o n, it \.·;111 "'c , h,'"" 1~ rliat
llic prnhahil it)' 1 11' t "''ti ,
Binomial J i:--trihut it,11 l·an Ix· '-·lt " cl ~ af'P'"' 1111 •1.1l'd hy lhc prnhahil, tv ,,r 1 f •,r.,.
' lie '
in Poi s~()ll di,t ri f, ut ion \\ ith r ar;111wtcr /11 ::.= ,,,, . l. l' . • l 1• •r
/P l I
(' fl I

.r !

I -
II ( ,, - I )( II ·- ~ -' _.. --~
- --- - -- -----
~-~ - ~~ , ,·, I
Proof .\ '

,,, r ( I
111
---
II j

tend~ to J and

,I, ,__m )ft - I

tend, t,) the limitin g value e-"'. Thus, under the condi tions stated above
\ n
I
"C
.,\
,-,rqn - x = -
X !
l l, 1. I, .. . 1] m '( e _,,,

= e-111.
\' I

Since th e prubabilitie~ of Poisson dis tribution are easier to compute than those of
Binomial di'.-ltribut ion. fo r practical purposes the Poisson approximation is used when
p is Jess than 0.1 and np is not very large (say, less than I 0).
Poisson di stri bution thus funds important application in such phenomenon where
the probab ility of occurrence of an event (success) is extremely smaJI, but the num~r
of oppor1uni tie~ is infi ni Lely large, so that the mean is something fini te. Hence thi,
di \ tri bu tion is ~ometimes known as the ..di stribution of rare events''.
Example 12.20 In turning ow certain toys in a manufacturing process in afi,d(''r.
till, m eruge numbn of de_(ecriw1 il 10'} . What is the prohabili(\' o/~£·tti11~ t'.rnrt1' J
Jt~/erri1·eJ in a sample of / 0 wys clwsen at m11do111, by usinx tht' Poisson ap11m.\i11:u!l:'1:
1
io the binomial distribi11iun Tak e r :::: 1. 72). _"J ( 11.C. \.\'./\ .. lk r . · ·~
Solution Lt:t the (x;currenct: <11 :1 dt ll'.( t iH: 1oy he c illl:d J .. ~uc:c\.' ,'.-1 ··. The nurnl·l\.'I , d J1.-l~1.·li\t''
1, I to! J,-,1A \ binnrrn.:1I Ji~IIi butiun with p~1, ,ti l ll' !\.' i .._ 11 :-: Io anJ p =- I {)I ~ -:. o I JI Lhi, d1, 11·ih ;t:d 1
J, tu be c1ppn1,; imatt:d b) J-'qj,; ,._ ,,11 d"1nhu1 1,n1
1:,
f I .\ I - ,, ,,,
I •

" e ha ve 111 == 111 1 -= 1<1 , I , .'

i (.
J I
- (1 ,,. 2.n c.- 1)_(, f

- ,, binomiJI Jislributio11 . the lntc r,·• -- 1 .


. tl'' l 1~ 111 C 11 •U,llll 111 y I\ 11 ,, . ( () I . 1 7 •
~o ·
l . (: \'Cll. f0 r 'a ·small 11 = 10 • thi -. '•,,•r•·· - • 1 · 1 10. J1 I - fl ~! l u p TP ~
c. ... t: :-. 1:..i11 1Y\\cll , h ,
._., 111::il~- - , •ing only .00-l) .I · \\ii lhl· I ur -. •.1111 ;.i pprnw.1 mav1on llfil
'1( j' 1rcn:nn: 1'I\'.
ithl' i'
ITII
le 12.21 2<7i of th<· iT l' 111 ., nwr/c h . 11111 ·/ . . ..
Esa01P . 111 . . , . . . . ·' " < 11111· ari, ,1£·/t<1n f ,nd th<• 1( ' .
·t·r.· rh1 1f 3 mc ,r, Ill\ at<' d1'/fc111 •,, in 1 ,.
' ,mn1 1I' · o/ I 0() ,rem.
·
, f(,1•·vn1 l' 1 -
0 1
. ,J,11 I111 • 1 , ·
11 049
1' l -2 - () J 15. l' =' l'l). II)'
f)_.,t'>~- c - · - 1p. Management ' 771
. i/11tiOll lhc nu'.nbc'.· nf defective~_(r) follow\ hi1101111 ..d di , 1rihu11011 "iir1l l' /> -: 2'¼ ~ 0.2 i,
)C d - JOO 1s f.urly large. mak111° np - ton (J> •
,n1,1 II. :in 1~ - • • • . • . e- _ - x · ~ = 2 a l1n1!L' quan11 1y. v,c u•.c Poi -, ..,,n
· · ,nuon to bmomral d1 stnhut1on with m = "fl == J
aprr1~\ ll • -·
m' 2'
f (_r ) = I' .'II , -:: (' ~ -
X ! .,\ !
Probahilily of 3 or more defecti ves
= /'(:,) +.f(4) + f(5) + .. . ..
= 1 - [f(0}+f(l ) +JOJ I

=I -,-, [1 +2+ t,]


= I - 0.135 x 5
= 0.325

§ HYPERGEOMETRI C DISTRIBUTION
Suppose that a box contains N balls among which A balls me white ;mJ the- renuin ing
N-A are black . If n ball s arc drawn at random without replacement , the prohahilil~
ofobtaining x white balls (and oh1viousl y 11 ~ x hlack ball,) among them i, gih:n by
AC ' \C
f( x ) = ' ., ., :( x = 0. 1.2 ... .. 111) , !2 .:- . li
\' c I

"where mis tl 1e smaller of the po,itin: iute.~Lr, 11 ,11HI .\ . ThL pn,t,Jbility d1,1nbut11,n
de_fined by the prohability ma"" fu ncll m: ( l .~.X I 1" ~1ww n :i, "' !'·, . _.( .,,ict , i,· 1 1

01 Hr1bu rion.

1· It may he nuted that (1 2.ti . l I sa1 i,tit· -.. h111ti tlw L' lllhii ti nn, t',11 :i ,,.m._f \i/ .
l/J(q>
. · - Q·, becau-..e the numerator and d~nomrn~11t11 · . :tr..., Jll > 1,·,,· · i11t• ,• r , · (1 1 1 ' 11 \ 1 : . 11
, \ \ =- · - ·
~-htt:ause the LI I. S. rqxescnls tht· , um of probahilit it's fnr all mutu:tlh n,h1,1, :! \
11n~'il'X
i hausti ve C\'(:ll l '' ,.\ = U. I. ')
-· .. .. "' .
1 .
lt: relation (ii ) m:1y abo he obt~ined by u-..ing tlw ,drnt it ~
1•1

Y
L-;
·\c· , \· •'c ,,.,
~ ,·c 11
.,
'f . f • \ ·, • • \ '' ( ',
i I ' .+ .

You might also like